---
格式版本: 2
标题: "Hybrid-LUT: Channel-Aware Hybrid Lookup Table and Filtering for Efficient Image Denoising"
原文链接: "https://arxiv.org/abs/2608.11646"
发布日期: "2026-08-12"
发布时间校准状态: "found"
发布时间需复核: "否"
发布时间来源: "rule:local:strict_original_body"
发布时间证据: "**\\[v1\\]** Wed, 12 Aug 2026 04:45:05 UTC (3,420 KB)"
发布时间校准原因: "规则确认唯一严格发布时间，来源 local:strict_original_body"
发布时间校准置信度: "high"
发布时间候选数量: 18
发布时间严格候选数量: 6
发布时间原页读取状态: "source template page reused from URL open"
发布时间未找到原因: ""
发布时间校准时间: "2026-08-13T18:34:33+08:00"
发布时间仲裁状态: "skipped"
发布时间仲裁尝试次数: 0
发布时间仲裁耗时毫秒: 0
发现时间: "2026-08-13T18:24:32+08:00"
入库时间: "2026-08-13T10:34:33.248Z"
来源平台: "arXiv 学术论文搜索"
搜索渠道: "source_template"
搜索词: "https://arxiv.org/search/?query=AI&searchtype=all"
匹配关键词:
  - "AI"
  - "SRAM"
  - "performance"
  - "throughput"
相关厂家:
  []
相关专家:
  []
内容类型: "网页"
抓取工具: "Free Fetch + Defuddle"
清洗工具: "Defuddle Markdown + Defuddle/Readability 正文提取"
原始附件:
  []
AI质检状态: "评分失败"
AI评分尝试次数: 1
AI评分错误类型: "service_error"
AI评分错误: "LLM call failed; tried model chain: ali-deepseek-v4-flash -> tx-deepseek-v4-flash | Model ali-deepseek-v4-flash failed 503: {\"error\":{\"code\":\"model_not_found\",\"message\":\"No available channel for model ali-deepseek-v4-flash under group vip (distributor) (request id: 202608131034364323969968268d9d6iiczP9NU)\",\"type\":\"new_api_error\"}} | Model tx-deepseek-v4-flash failed 503: {\"error\":{\"code\":\"model_not_found\",\"message\":\"No available channel for model tx-deepseek-v4-flash under group vip (distributor"
AI评分开始时间: "2026-08-13T10:34:33.255Z"
AI评分结束时间: "2026-08-13T10:34:37.074Z"
AI摘要: "Hybrid-LUT 提出一种基于 YUV 的非对称图像去噪框架，Y 通道用多波段查找表恢复纹理，UV 通道用轻量滤波保持色彩，相比 RGB-LUT 方法将查找表存储减少三分之二，总存储仅 421 KB；"
AI摘要模型: "ali-deepseek-v4-flash"
AI摘要时间: "2026-09-07T03:40:24.652Z"
采集批次: "2026年8月13日18点24分27秒"
采集批次ID: "20260813-182427-1564e572"
去重键: "https://arxiv.org/abs/2608.11646"
---

## Computer Science > Computer Vision and Pattern Recognition

## Title:Hybrid-LUT: Channel-Aware Hybrid Lookup Table and Filtering for Efficient Image Denoising

Authors:[Zhilin Ai](https://arxiv.org/search/cs?searchtype=author&query=Ai,+Z), [Boyu Li](https://arxiv.org/search/cs?searchtype=author&query=Li,+B), [Sidi Yang](https://arxiv.org/search/cs?searchtype=author&query=Yang,+S), [Wenqing Shi](https://arxiv.org/search/cs?searchtype=author&query=Shi,+W), [Wenyong Zhou](https://arxiv.org/search/cs?searchtype=author&query=Zhou,+W), [Binxiao Huang](https://arxiv.org/search/cs?searchtype=author&query=Huang,+B), [Chenchen Ding](https://arxiv.org/search/cs?searchtype=author&query=Ding,+C), [Ngai Wong](https://arxiv.org/search/cs?searchtype=author&query=Wong,+N)

[View PDF](https://arxiv.org/pdf/2608.11646) [HTML (experimental)](https://arxiv.org/html/2608.11646v1)

> Abstract:Lookup table (LUT)-based image denoising methods have attracted increasing attention due to their high efficiency and hardware-friendly properties. However, existing RGB-LUT approaches require three identical LUTs to process RGB channels in parallel, resulting in large on-chip SRAM consumption. A simple alternative is to apply LUT processing only to the luminance (Y) channel in the YUV color space to reduce memory usage. However, this naive strategy leads to degraded restoration quality, since ignoring the chrominance (UV) channels introduces color distortion and residual artifacts. In this work, we propose Hybrid-LUT, a YUV-based asymmetric channel-processing framework that combines LUT and filtering in a unified design. Specifically, a multi-band LUT branch with pixel-level weight fusion is applied to the Y channel to recover fine textures, while lightweight filtering is used for the UV channels to maintain color consistency. This design reduces LUT storage by two-thirds compared with RGB-LUT methods while maintaining the same runtime throughput. Extensive experiments show that Hybrid-LUT achieves state-of-the-art (SOTA) performance across multiple benchmarks with only 421 KB of storage. In particular, our method surpasses existing LUT-based denoising approaches by at least 0.63 dB CPSNR on real-world datasets, demonstrating its effectiveness for image denoising on resource-constrained edge devices. The project is available at [this https URL](https://github.com/Ai-ZL/Hybrid-LUT).

| Comments: |  |
| --- | --- |
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Image and Video Processing (eess.IV) |
| Cite as: | [arXiv:2608.11646](https://arxiv.org/abs/2608.11646) \[cs.CV\] |
|  | (or [arXiv:2608.11646v1](https://arxiv.org/abs/2608.11646v1) \[cs.CV\] for this version) |
|  | [https://doi.org/10.48550/arXiv.2608.11646](https://doi.org/10.48550/arXiv.2608.11646) |

## Submission history

From: Zhilin Ai \[[view email](https://arxiv.org/show-email/ad0e55ca/2608.11646)\]  
**\[v1\]** Wed, 12 Aug 2026 04:45:05 UTC (3,420 KB)

[Which authors of this paper are endorsers?](https://arxiv.org/auth/show-endorsers/2608.11646) | Disable MathJax ([What is MathJax?](https://info.arxiv.org/help/mathjax.html))